Christian Huitema
Interests: Books: Information:

IPv6

I have been working on IPv6 in various capacities since 1992. In 1995, I published a book describing the state of the just-adopted standard, and the rationales behind the design choices. More recently, at Microsoft, I have worked on transition technologies such as 6to4 and Teredo.

IPv6, the new Internet protocol

This book describes the new Internet Protocol, IPv6. Two editions have been produced. The first edition describes the state of the IPv6 work in 1995. Its main value is now historic. You can check here the reference page on the Prentice hall web site. Things progressed fast, and a second edition tried to capture the state of IPv6 standardization in 1997. There have been miscellaneous changes since 1997, notably the definition of the 6to4 address format that facilitates transition. Another change introduced in 1998, the A6 DNS format, has since been rolled back. After that, the main areas of innovation have been the development of new transition technologies, notably 6to4, ISATAP and Teredo. An IETF working group is also working on the SHIM6 proposal to support multihoming. Of course, none of that is described in the 1995 or 1997 editions.

How log until we need to deploy IPv6 ?

The running joke in the Internet community was that IPv6 will be deployed "in the next 10 years." Indeed, more than 10 years have now passed since the protocol was standardized. There is some usage on the Internet, but IPv6 has certainly not yet displaced IPv4. In December 2000, I wrote a page trying to answer the question, How long can we wait before we need IPv6? The basic argument developed in the page was as the Internet grows, we will at some point run out of IPv4 addresses. My prediction at the time was that we would run out of addresses by 2009. It is always interesting to come back and revisit these old predictions...

Can we actually deploy IPv6?

Many networking specialists doubt that we will ever deploy IPv6. Their doubts are rooted in the huge cost of a network transition. Introducing a new network protocol requires changing about every component of the Internet, or at a minimum rewriting their software or firmware. They don't believe that the technical advantages of IPv6 are sufficient to justify such a costly makeover.

I personally believe that the cost of the transition can be minimized by the judicious choice of transition technologies, and by the advantage that the continuous IPv6 addressing provides for peer-to-peer applications, including VoIP.